diplodocus.org Git - nmh/blob - uip/sortm.c

   1 /* sortm.c -- sort messages in a folder by date/time
   2  *
   3  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   4  * COPYRIGHT file in the root directory of the nmh distribution for
   5  * complete copyright information.
   6  */
   7
   8 #include "h/mh.h"
   9 #include "sbr/dtime.h"
  10 #include "sbr/m_name.h"
  11 #include "sbr/m_getfld.h"
  12 #include "sbr/getarguments.h"
  13 #include "sbr/seq_setprev.h"
  14 #include "sbr/seq_setcur.h"
  15 #include "sbr/seq_save.h"
  16 #include "sbr/smatch.h"
  17 #include "sbr/uprf.h"
  18 #include "sbr/m_convert.h"
  19 #include "sbr/getfolder.h"
  20 #include "sbr/ext_hook.h"
  21 #include "sbr/folder_read.h"
  22 #include "sbr/folder_free.h"
  23 #include "sbr/context_save.h"
  24 #include "sbr/context_replace.h"
  25 #include "sbr/context_find.h"
  26 #include "sbr/ambigsw.h"
  27 #include "sbr/path.h"
  28 #include "sbr/print_version.h"
  29 #include "sbr/print_help.h"
  30 #include "sbr/error.h"
  31 #include "h/tws.h"
  32 #include "h/done.h"
  33 #include "h/utils.h"
  34 #include "sbr/m_maildir.h"
  35
  36 #define SORTM_SWITCHES \
  37     X("datefield field", 0, DATESW) \
  38     X("textfield field", 0, TEXTSW) \
  39     X("notextfield", 0, NSUBJSW) \
  40     X("subject", -3, SUBJSW) /* backward-compatibility */ \
  41     X("limit days", 0, LIMSW) \
  42     X("nolimit", 0, NLIMSW) \
  43     X("verbose", 0, VERBSW) \
  44     X("noverbose", 0, NVERBSW) \
  45     X("all", 0, ALLMSGS) \
  46     X("noall", 0, NALLMSGS) \
  47     X("check", 0, CHECKSW) \
  48     X("nocheck", 0, NCHECKSW) \
  49     X("version", 0, VERSIONSW) \
  50     X("help", 0, HELPSW) \
  51
  52 #define X(sw, minchars, id) id,
  53 DEFINE_SWITCH_ENUM(SORTM);
  54 #undef X
  55
  56 #define X(sw, minchars, id) { sw, minchars, id },
  57 DEFINE_SWITCH_ARRAY(SORTM, switches);
  58 #undef X
  59
  60 struct smsg {
  61     int s_msg;
  62     time_t s_clock;
  63     char *s_subj;
  64 };
  65
  66 static struct smsg *smsgs;
  67 int nmsgs;
  68
  69 char *subjsort;                 /* sort on subject if != 0 */
  70 time_t datelimit = 0;
  71 bool submajor;                  /* if true, sort on subject-major */
  72 bool verbose;
  73 int allmsgs = 1;
  74 int check_failed = 0;
  75
  76 /* This keeps compiler happy on calls to qsort */
  77 typedef int (*qsort_comp) (const void *, const void *);
  78
  79 /*
  80  * static prototypes
  81  */
  82 static int read_hdrs (struct msgs *, char *);
  83 static int get_fields (char *, int, struct smsg *);
  84 static int dsort (struct smsg **, struct smsg **);
  85 static int subsort (struct smsg **, struct smsg **);
  86 static int txtsort (struct smsg **, struct smsg **);
  87 static void rename_chain (struct msgs *, struct smsg **, int, int);
  88 static void rename_msgs (struct msgs *, struct smsg **);
  89
  90
  91 int
  92 main (int argc, char **argv)
  93 {
  94     int i, msgnum;
  95     char *cp, *maildir, *datesw = NULL;
  96     char *folder = NULL, buf[BUFSIZ], **argp;
  97     char **arguments;
  98     struct msgs_array msgs = { 0, 0, NULL };
  99     struct msgs *mp;
 100     struct smsg **dlist;
 101     bool checksw = false;
 102
 103     if (nmh_init(argv[0], true, true)) { return 1; }
 104
 105     arguments = getarguments (invo_name, argc, argv, 1);
 106     argp = arguments;
 107
 108     /*
 109      * Parse arguments
 110      */
 111     while ((cp = *argp++)) {
 112         if (*cp == '-') {
 113             switch (smatch (++cp, switches)) {
 114             case AMBIGSW:
 115                 ambigsw (cp, switches);
 116                 done (1);
 117             case UNKWNSW:
 118                 die("-%s unknown", cp);
 119
 120             case HELPSW:
 121                 snprintf(buf, sizeof(buf), "%s [+folder] [msgs] [switches]",
 122                         invo_name);
 123                 print_help (buf, switches, 1);
 124                 done (0);
 125             case VERSIONSW:
 126                 print_version(invo_name);
 127                 done (0);
 128
 129             case DATESW:
 130                 if (datesw)
 131                     die("only one date field at a time");
 132                 if (!(datesw = *argp++) || *datesw == '-')
 133                     die("missing argument to %s", argp[-2]);
 134                 continue;
 135
 136             case TEXTSW:
 137                 if (subjsort)
 138                     die("only one text field at a time");
 139                 if (!(subjsort = *argp++) || *subjsort == '-')
 140                     die("missing argument to %s", argp[-2]);
 141                 continue;
 142
 143             case SUBJSW:
 144                 subjsort = "subject";
 145                 continue;
 146             case NSUBJSW:
 147                 subjsort = NULL;
 148                 continue;
 149
 150             case LIMSW:
 151                 if (!(cp = *argp++) || *cp == '-')
 152                         die("missing argument to %s", argp[-2]);
 153                 while (*cp == '0')
 154                     cp++;               /* skip any leading zeros */
 155                 if (!*cp) {             /* hit end of string */
 156                     submajor = true;    /* sort subject-major */
 157                     continue;
 158                 }
 159                 if (!isdigit((unsigned char) *cp) || !(datelimit = atoi(cp)))
 160                     die("impossible limit %s", cp);
 161                 datelimit *= 60*60*24;
 162                 continue;
 163             case NLIMSW:
 164                 submajor = false;       /* use date-major, but */
 165                 datelimit = 0;          /* use no limit */
 166                 continue;
 167
 168             case VERBSW:
 169                 verbose = true;
 170                 continue;
 171             case NVERBSW:
 172                 verbose = false;
 173                 continue;
 174
 175             case ALLMSGS:
 176                 allmsgs = 1;
 177                 continue;
 178             case NALLMSGS:
 179                 allmsgs = 0;
 180                 continue;
 181
 182             case CHECKSW:
 183                 checksw = true;
 184                 continue;
 185             case NCHECKSW:
 186                 checksw = false;
 187                 continue;
 188             }
 189         }
 190         if (*cp == '+' || *cp == '@') {
 191             if (folder)
 192                 die("only one folder at a time!");
 193             folder = pluspath (cp);
 194         } else
 195             app_msgarg(&msgs, cp);
 196     }
 197
 198     if (!context_find ("path"))
 199         free (path ("./", TFOLDER));
 200     if (!msgs.size) {
 201         if (allmsgs) {
 202             app_msgarg(&msgs, "all");
 203         } else {
 204             die("must specify messages to sort with -noall");
 205         }
 206     }
 207     if (!datesw)
 208         datesw = "date";
 209     if (!folder)
 210         folder = getfolder (1);
 211     maildir = m_maildir (folder);
 212
 213     if (chdir (maildir) == NOTOK)
 214         adios (maildir, "unable to change directory to");
 215
 216     /* read folder and create message structure */
 217     if (!(mp = folder_read (folder, 1)))
 218         die("unable to read folder %s", folder);
 219
 220     /* check for empty folder */
 221     if (mp->nummsg == 0)
 222         die("no messages in %s", folder);
 223
 224     /* parse all the message ranges/sequences and set SELECTED */
 225     for (msgnum = 0; msgnum < msgs.size; msgnum++)
 226         if (!m_convert (mp, msgs.msgs[msgnum]))
 227             done (1);
 228     seq_setprev (mp);   /* set the previous sequence */
 229
 230     if ((nmsgs = read_hdrs (mp, datesw)) <= 0)
 231         die("no messages to sort");
 232
 233     if (checksw  &&  check_failed) {
 234         die("errors found, no messages sorted");
 235     }
 236
 237     /*
 238      * sort a list of pointers to our "messages to be sorted".
 239      */
 240     dlist = mh_xmalloc ((nmsgs+1) * sizeof(*dlist));
 241     for (i = 0; i < nmsgs; i++)
 242         dlist[i] = &smsgs[i];
 243     dlist[nmsgs] = 0;
 244
 245     if (verbose) {      /* announce what we're doing */
 246         if (subjsort)
 247             if (submajor)
 248                 printf ("sorting by %s\n", subjsort);
 249             else
 250                 printf ("sorting by %s-major %s-minor\n", subjsort, datesw);
 251         else
 252             printf ("sorting by datefield %s\n", datesw);
 253     }
 254
 255     /* first sort by date, or by subject-major, date-minor */
 256     qsort (dlist, nmsgs, sizeof(*dlist),
 257             (qsort_comp) (submajor && subjsort ? txtsort : dsort));
 258
 259     /*
 260      * if we're sorting on subject, we need another list
 261      * in subject order, then a merge pass to collate the
 262      * two sorts.
 263      */
 264     if (!submajor && subjsort) {        /* already date sorted */
 265         struct smsg **slist, **flist;
 266         struct smsg ***il, **fp, **dp;
 267
 268         slist = mh_xmalloc ((nmsgs+1) * sizeof(*slist));
 269         memcpy(slist, dlist, (nmsgs+1)*sizeof(*slist));
 270         qsort(slist, nmsgs, sizeof(*slist), (qsort_comp) subsort);
 271
 272         /*
 273          * make an inversion list so we can quickly find
 274          * the collection of messages with the same subj
 275          * given a message number.
 276          */
 277         il = mh_xcalloc(mp->hghsel + 1, sizeof *il);
 278         for (i = 0; i < nmsgs; i++)
 279             il[slist[i]->s_msg] = &slist[i];
 280         /*
 281          * make up the final list, chronological but with
 282          * all the same subjects grouped together.
 283          */
 284         flist = mh_xmalloc ((nmsgs+1) * sizeof(*flist));
 285         fp = flist;
 286         for (dp = dlist; *dp;) {
 287             struct smsg **s = il[(*dp++)->s_msg];
 288
 289             /* see if we already did this guy */
 290             if (! s)
 291                 continue;
 292
 293             *fp++ = *s++;
 294             /*
 295              * take the next message(s) if there is one,
 296              * its subject isn't null and its subject
 297              * is the same as this one and it's not too
 298              * far away in time.
 299              */
 300             while (*s && (*s)->s_subj[0] &&
 301                    strcmp((*s)->s_subj, s[-1]->s_subj) == 0 &&
 302                    (datelimit == 0 ||
 303                    (*s)->s_clock - s[-1]->s_clock <= datelimit)) {
 304                 il[(*s)->s_msg] = 0;
 305                 *fp++ = *s++;
 306             }
 307         }
 308         *fp = 0;
 309         free (il);
 310         free (slist);
 311         free (dlist);
 312         dlist = flist;
 313     }
 314
 315     /*
 316      * At this point, dlist is a sorted array of pointers to smsg structures,
 317      * each of which contains a message number.
 318      */
 319
 320     rename_msgs (mp, dlist);
 321
 322     context_replace (pfolder, folder);  /* update current folder         */
 323     seq_save (mp);                      /* synchronize message sequences */
 324     context_save ();                    /* save the context file         */
 325     folder_free (mp);                   /* free folder/message structure */
 326     done (0);
 327     return 1;
 328 }
 329
 330 static int
 331 read_hdrs (struct msgs *mp, char *datesw)
 332 {
 333     int msgnum;
 334     struct smsg *s;
 335
 336     smsgs = mh_xcalloc(mp->hghsel - mp->lowsel + 2, sizeof *smsgs);
 337     s = smsgs;
 338     for (msgnum = mp->lowsel; msgnum <= mp->hghsel; msgnum++) {
 339         if (is_selected(mp, msgnum)) {
 340             if (get_fields (datesw, msgnum, s)) {
 341                 s->s_msg = msgnum;
 342                 s++;
 343             }
 344         }
 345     }
 346     s->s_msg = 0;
 347     return s - smsgs;
 348 }
 349
 350
 351 /*
 352  * Parse the message and get the data or subject field,
 353  * if needed.
 354  */
 355
 356 static int
 357 get_fields (char *datesw, int msg, struct smsg *smsg)
 358 {
 359     int state;
 360     int compnum;
 361     char *msgnam, buf[NMH_BUFSIZ], nam[NAMESZ];
 362     struct tws *tw;
 363     char *datecomp = NULL, *subjcomp = NULL;
 364     FILE *in;
 365     m_getfld_state_t gstate;
 366
 367     if ((in = fopen (msgnam = m_name (msg), "r")) == NULL) {
 368         admonish (msgnam, "unable to read message");
 369         return 0;
 370     }
 371     gstate = m_getfld_state_init(in);
 372     for (compnum = 1;;) {
 373         int bufsz = sizeof buf;
 374         switch (state = m_getfld2(&gstate, nam, buf, &bufsz)) {
 375         case FLD:
 376         case FLDPLUS:
 377             compnum++;
 378             if (!strcasecmp (nam, datesw)) {
 379                 datecomp = add (buf, datecomp);
 380                 while (state == FLDPLUS) {
 381                     bufsz = sizeof buf;
 382                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 383                     datecomp = add (buf, datecomp);
 384                 }
 385                 if (!subjsort || subjcomp)
 386                     break;
 387             } else if (subjsort && !strcasecmp (nam, subjsort)) {
 388                 subjcomp = add (buf, subjcomp);
 389                 while (state == FLDPLUS) {
 390                     bufsz = sizeof buf;
 391                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 392                     subjcomp = add (buf, subjcomp);
 393                 }
 394                 if (datecomp)
 395                     break;
 396             } else {
 397                 /* just flush this guy */
 398                 while (state == FLDPLUS) {
 399                     bufsz = sizeof buf;
 400                     state = m_getfld2(&gstate, nam, buf, &bufsz);
 401                 }
 402             }
 403             continue;
 404
 405         case BODY:
 406         case FILEEOF:
 407             break;
 408
 409         case LENERR:
 410         case FMTERR:
 411             if (state == LENERR || state == FMTERR) {
 412                 inform("format error in message %d (header #%d), continuing...",
 413                       msg, compnum);
 414                 check_failed = 1;
 415             }
 416             free(datecomp);
 417             free(subjcomp);
 418             fclose (in);
 419             return 0;
 420
 421         default:
 422             die("internal error -- you lose");
 423         }
 424         break;
 425     }
 426     m_getfld_state_destroy (&gstate);
 427
 428     /*
 429      * If no date component, then use the modification
 430      * time of the file as its date
 431      */
 432     if (!datecomp || (tw = dparsetime (datecomp)) == NULL) {
 433         struct stat st;
 434
 435         inform("can't parse %s field in message %d, "
 436             "will use file modification time", datesw, msg);
 437         fstat (fileno (in), &st);
 438         smsg->s_clock = st.st_mtime;
 439         check_failed = 1;
 440     } else {
 441         smsg->s_clock = dmktime (tw);
 442     }
 443
 444     if (subjsort) {
 445         if (subjcomp) {
 446             /*
 447              * try to make the subject "canonical": delete
 448              * leading "re:", everything but letters & smash
 449              * letters to lower case.
 450              */
 451             char  *cp, *cp2, c;
 452
 453             cp = subjcomp;
 454             cp2 = subjcomp;
 455             if (strcmp (subjsort, "subject") == 0) {
 456                 while ((c = *cp)) {
 457                     if (! isspace((unsigned char) c)) {
 458                         if(!uprf(cp, "re:"))
 459                             break;
 460                         cp += 2;
 461                     }
 462                     cp++;
 463                 }
 464             }
 465
 466             while ((c = *cp++)) {
 467                 if (isascii((unsigned char) c) && isalnum((unsigned char) c))
 468                     *cp2++ = tolower((unsigned char)c);
 469             }
 470
 471             *cp2 = '\0';
 472         }
 473         else
 474             subjcomp = "";
 475
 476         smsg->s_subj = subjcomp;
 477     }
 478     fclose (in);
 479     free(datecomp);
 480
 481     return 1;
 482 }
 483
 484 /*
 485  * sort on dates.
 486  */
 487 static int
 488 dsort (struct smsg **a, struct smsg **b)
 489 {
 490     if ((*a)->s_clock < (*b)->s_clock)
 491         return -1;
 492     if ((*a)->s_clock > (*b)->s_clock)
 493         return 1;
 494     if ((*a)->s_msg < (*b)->s_msg)
 495         return -1;
 496     return 1;
 497 }
 498
 499 /*
 500  * sort on subjects.
 501  */
 502 static int
 503 subsort (struct smsg **a, struct smsg **b)
 504 {
 505     int i;
 506
 507     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 508         return i;
 509
 510     return dsort(a, b);
 511 }
 512
 513 static int
 514 txtsort (struct smsg **a, struct smsg **b)
 515 {
 516     int i;
 517
 518     if ((i = strcmp ((*a)->s_subj, (*b)->s_subj)))
 519         return i;
 520     if ((*a)->s_msg < (*b)->s_msg)
 521         return -1;
 522     return 1;
 523 }
 524
 525 static void
 526 rename_chain (struct msgs *mp, struct smsg **mlist, int msg, int endmsg)
 527 {
 528     int nxt, old, new;
 529     char *newname, oldname[BUFSIZ];
 530     char newbuf[PATH_MAX + 1];
 531
 532     for (;;) {
 533         nxt = mlist[msg] - smsgs;       /* mlist[msg] is a ptr into smsgs */
 534         mlist[msg] = NULL;
 535         old = smsgs[nxt].s_msg;
 536         new = smsgs[msg].s_msg;
 537         strncpy (oldname, m_name (old), sizeof(oldname));
 538         newname = m_name (new);
 539         if (verbose)
 540             printf ("message %d becomes message %d\n", old, new);
 541
 542         (void)snprintf(oldname, sizeof (oldname), "%s/%d", mp->foldpath, old);
 543         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, new);
 544         ext_hook("ref-hook", oldname, newbuf);
 545
 546         if (rename (oldname, newname) == NOTOK)
 547             adios (newname, "unable to rename %s to", oldname);
 548
 549         copy_msg_flags (mp, new, old);
 550         if (mp->curmsg == old)
 551             seq_setcur (mp, new);
 552
 553         if (nxt == endmsg)
 554             break;
 555
 556         msg = nxt;
 557     }
 558 /*      if (nxt != endmsg); */
 559 /*      rename_chain (mp, mlist, nxt, endmsg); */
 560 }
 561
 562 static void
 563 rename_msgs (struct msgs *mp, struct smsg **mlist)
 564 {
 565     int i, j, old, new;
 566     bvector_t tmpset = bvector_create ();
 567     char f1[BUFSIZ], tmpfil[BUFSIZ];
 568     char newbuf[PATH_MAX + 1];
 569     struct smsg *sp;
 570
 571     strncpy (tmpfil, m_name (mp->hghmsg + 1), sizeof(tmpfil));
 572
 573     for (i = 0; i < nmsgs; i++) {
 574         if (! (sp = mlist[i]))
 575             continue;   /* did this one */
 576
 577         j = sp - smsgs;
 578         if (j == i)
 579             continue;   /* this one doesn't move */
 580
 581         /*
 582          * the guy that was msg j is about to become msg i.
 583          * rename 'j' to make a hole, then recursively rename
 584          * guys to fill up the hole.
 585          */
 586         old = smsgs[j].s_msg;
 587         new = smsgs[i].s_msg;
 588         strncpy (f1, m_name (old), sizeof(f1));
 589
 590         if (verbose)
 591             printf ("renaming message chain from %d to %d\n", old, new);
 592
 593         /*
 594          *      Run the external hook to refile the old message as the
 595          *      temporary message number that is off of the end of the
 596          *      messages in the folder.
 597          */
 598
 599         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, old);
 600         (void)snprintf(newbuf, sizeof (newbuf), "%s/%d", mp->foldpath, mp->hghmsg + 1);
 601         ext_hook("ref-hook", f1, newbuf);
 602
 603         if (rename (f1, tmpfil) == NOTOK)
 604             adios (tmpfil, "unable to rename %s to ", f1);
 605
 606         get_msg_flags (mp, tmpset, old);
 607
 608         rename_chain (mp, mlist, j, i);
 609
 610         /*
 611          *      Run the external hook to refile the temporary message number
 612          *      to the real place.
 613          */
 614
 615         (void)snprintf(f1, sizeof (f1), "%s/%d", mp->foldpath, new);
 616         ext_hook("ref-hook", newbuf, f1);
 617
 618         if (rename (tmpfil, m_name(new)) == NOTOK)
 619             adios (m_name(new), "unable to rename %s to", tmpfil);
 620
 621         set_msg_flags (mp, tmpset, new);
 622         mp->msgflags |= SEQMOD;
 623     }
 624
 625     bvector_free (tmpset);
 626 }